Coldcard Wallet Flaw Drains 594 BTC Worth $38M in 25-Minute Sweep

A 25-Minute Sweep Hits Bitcoin’s Most Trusted Cold Storage

Around 594 Bitcoin, worth roughly $38 million, was drained from about 500 hardware wallets in a coordinated sweep that lasted just 25 minutes early on Friday. The attack ran between 01:31 and 01:56 UTC, moving 1,324 chunks of Bitcoin across 500 separate transactions inside a three-block window, according to CoinDesk.

The stolen coins were consolidated into a single address (bc1qq85v2c926eg6pgxhwp6q7lf6cnsz80qs3fcu9r), which held 562 BTC as of publication and has not yet moved. Every drained wallet held more than 0.15 BTC and had sat inactive for extended periods, some dating back to 2021.

The Firmware Bug That Broke Cold Storage

The culprit is a subtle failure in how Coldcard Mk3 devices generated the random numbers that secure Bitcoin seeds. Instead of using the dedicated hardware randomness chip built for the job, affected devices quietly fell back to a software routine seeded by non-secret data such as the device serial number and clock registers.

That produced seed phrases with only about 72 bits of entropy, well below the 128 bits considered the industry floor for a 12-word BIP-39 mnemonic. Once the pattern was understood, the pool of possible seeds became small enough for attackers to enumerate. Block’s Bitcoin engineering team disclosed the findings to Coinkite, the Canadian company behind Coldcard, before publishing.

Which Coldcard Devices Are Affected

Coinkite confirmed that the vulnerability was introduced in Coldcard firmware version 4.0.0, released in March 2021, and applied only to the Mk3 line running that version or later. Newer Coldcard hardware, including Mk4, Mk5, and the Q model, is not affected under Coinkite’s early analysis.

Beyond seed generation itself, the same faulty random number generator produced paper wallet private keys, seed splitting masks, device cloning keys, and Key Teleport transfers. Any Bitcoin secured through any of those paths on a vulnerable Mk3 is now considered exposed, even if the underlying digital wallet has never moved a satoshi.

How Seed Randomness Actually Works

Hardware wallets are designed to keep a user’s Bitcoin offline by generating a seed phrase inside the device, using a random source that cannot be predicted from outside. A digital wallet built on that seed can hold coins for years without ever touching the internet. If the randomness is weak, however, the seed becomes guessable.

Attackers can then enumerate possible seeds until one matches a known Bitcoin address on the public blockchain, then sweep the coins. That is what appears to have happened here. According to CryptoSlate, affected Mk3 devices generated seeds from a small, predictable pool, which was eventually mapped to real on-chain balances.

Coinkite Confirms, Block Publishes Under Pressure

Coinkite issued an urgent security advisory on July 30 stating that “Mk4, Q and Mk5 are not affected based on our early analysis.” Block chose to publish its research without full confirmation testing because active exploitation was already underway, an unusual step for a coordinated vulnerability disclosure. Both companies described their analyses as preliminary.

Owners of Mk3 devices have been told to move any remaining balances to a seed generated on a different device, ideally one from a separate manufacturer, and to treat every backup ever produced by a compromised Mk3 as burned. Coinkite has said a firmware update is in preparation, though it will not restore funds already stolen or reseed existing wallets.

Where the Stolen Bitcoin Went

The attacker consolidated 562 of the 594 stolen BTC into the single address bc1qq85v2c926eg6pgxhwp6q7lf6cnsz80qs3fcu9r, which has remained dormant since the sweep completed. On-chain analysts are watching the address for any movement toward exchanges or mixers, which could offer clues about the actor behind the theft.

The remaining 32 BTC was scattered across smaller addresses. Bitcoin was trading near $65,000 as the drain unfolded, and the broader bitcoin price held steady through the event, suggesting the market has largely absorbed the incident without a wider sell-off in the majors.

What Coldcard Owners Should Watch Next

Analysts studying the address cluster say further sweeps may still occur if the attackers hold a wider list of compromised seeds than they have exploited so far. The affected pool could extend beyond the 500 drained wallets, since the vulnerability window covers roughly five years of Mk3 use. Anyone who set up a Mk3 device between March 2021 and July 2026, and has not since migrated the funds, should treat their seed as compromised regardless of whether their balance has moved. Users tracking follow-up advisories and any second wave of activity can monitor the story through the latest crypto news coverage in the coming days.

A Trust Shock for the Hardware Wallet Category

Hardware wallets have long been marketed as the gold standard for self-custody, on the premise that a well-audited offline device eliminates the broader attack surface of hot wallets and centralised exchanges. The Coldcard incident does not break that thesis, but it does show how much of it rests on a single technical assumption: that the random number generator inside the device is doing what it says on the tin.

A five-year gap between the bug’s introduction and its exploitation also reveals how long a silent flaw can sit inside otherwise trusted hardware before adversaries catch up. For the self-custody community, the takeaway is neither to abandon hardware wallets nor to trust them blindly, but to treat entropy, firmware provenance, and cross-vendor diversification as first-order security decisions rather than afterthoughts.